Predicting Swiss Votes Through Machine Learning

Contact: Aswin Suresh

Switzerland follows a system of direct democracy where the citizens vote several times each year on laws and popular initiatives. Predicting the outcome of these votes is of interest to the public, politicians and media organisations.

The lab has an ongoing project Predikon that predicts the national outcome of the vote based on partial results from a few municipalities. The current project seeks to extend this to make predictions before the vote.

To make predictions we will rely on information sources that are available on the web before the vote, such as text from official voting booklets, news articles and social media, and metadata about votes and municipalities.

Requirements

  • Strong coding skills in Python
  • Experience in text mining and NLP a plus
  • Experience in Web scraping a plus
  • Interest in politics (and their mechanics) a plus

If interested, please send your CV and a transcript of your grades to aswin.suresh@epfl.ch.